© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 49152
    9. 8-RES_FS49152-JB
    11. 37290
    Prev
    Next
    INDEX
    Explanations

    the conclusion or ending statements in various contexts

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    Data Type
    torch.float32
    Hook Point
    blocks.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    459
    +0.12
    0.5%
    195
    +0.11
    0.5%
    276
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    276
    +0.12
    0.01
    702
    +0.11
    0.01
    492
    +0.10
    0.01
    Negative Logits
    ichick
    -0.69
    ategory
    -0.67
    alcohol
    -0.62
    terness
    -0.59
     Photographer
    -0.58
     absentee
    -0.58
     architect
    -0.58
    mingham
    -0.57
     oxid
    -0.57
     eleph
    -0.57
    POSITIVE LOGITS
    angered
    0.99
    urance
    0.98
    ragon
    0.94
    owment
    0.91
    orph
    0.90
    lich
    0.90
    ering
    0.90
    angering
    0.89
    ulum
    0.89
    orse
    0.85
    Activations Density 0.020%

    No Known Activations